Galactofuranose (Galf)-containing sugar chain contributes to the hyphal growth, conidiation and virulence of F. oxysporum f.sp. cucumerinum

The ascomycete fungus Fusarium oxysporum f.sp. cucumerinum causes vascular wilt diseases in cucumber. However, few genes related to morphogenesis and pathogenicity of this fungal pathogen have been functionally characterized. BLASTp searches of the Aspergillus fumigatus UgmA and galatofuranosyltransferases (Galf-transferases) sequences in the F. oxysporum genome identified two genes encoding putative UDP-galactopyranose mutase (UGM), ugmA and ugmB, and six genes encoding putative Galf-transferase homologs. In this study, the single and double mutants of the ugmA, ugmB and gfsB were obtained. The roles of UGMs and GfsB were investigated by analyzing the phenotypes of the mutants. Our results showed that deletion of the ugmA gene led to a reduced production of galactofuranose-containing sugar chains, reduced growth and impaired conidiation of F. oxysporum f.sp. cucumerinum. Most importantly, the ugmA deletion mutant lost the pathogenicity in cucumber plantlets. Although deletion of the ugmB gene did not cause any visible phenotype, deletion of both ugmA and ugmB genes caused more severe phenotypes as compared with the ΔugmA, suggesting that UgmA and UgmB are redundant and they can both contribute to synthesis of UDP-Galf. Furthermore, the ΔgfsB exhibited an attenuated virulence although no other phenotype was observed. Our results demonstrate that the galactofuranose (Galf) synthesis contributes to the cell wall integrity, germination, hyphal growth, conidiation and virulence in Fusarium oxysporum f.sp. cucumerinum and an ideal target for the development of new anti-Fusarium agents.

lipopolysaccharide (LPS) O-antigen domains, extracellular capsules, and polysaccharides [1]. In certain filamentous fungi, Galf is a major component of the cell wall and structural glycoproteins [2][3][4][5][6][7]. In pathogenic protozoa, Galf residues are key components of surface glycosylphosphoinositol (GPI)-anchored glycoconjugates in Leishmania and Trypanosoma cruzi [8][9][10]. In nematode such as Caenorhabditis elegans Galf residues are key components of surface coat [11]. In contrast, higher animals including humans and plants have no Galf-containing sugar chains. Therefore, inhibition of the biosynthetic pathway of Galf-containing sugar chains is an attractive target for the development of medicines and agricultural chemicals inhibiting pathogenic fungi, bacteria and parasites without any side effects. All Galf-containing sugar chains are synthesized using UDP-Galf as a sugar donor. The biosynthesis of UDP-Galf begins with the conversion of UDP-glucose to UDP-Galp by UDP-glucose 4-epimerase (UGE), then UDP-Galp is converted to UDP-Galf by UDP-galactopyranose mutase (UGM), and eventually Galf is transfered from UDP-Galf to Galf-containing sugar chains by galactofuranosyl(Galf)-transferases. In fungi, the site of Galf-containing sugar chains biosynthesis is the internal lumen of the Golgi apparatus. Therefore, a UDP-Galf transporter is necessary to transport the synthesized UDP-Galf from the cytosol to the inner lumen of the Golgi apparatus [12].
The genes responsible for the Galf biosynthesis have been identified in Aspergillus fumigatus, A. nidulans and A. niger, which include a similar gene sets of UDP-glucose 4-epimerases (UgeA), UDP-galactomutases (UgmA), UDP-Galf-transporters (UgtA and UgtB), and Galftransferases [13][14][15][16][17][18][19][20][21][22][23][24]. In A. nidulans, deletion of the ugeA, the gene encoding UDP-glucose 4-epimerase, leads to highly branched hyphae and reduced conidiation [15]. The A. niger ΔugmA strain is highly sensitive to the chitin binding agent calcofluor white, suggesting that synthesis of a Galf-containing sugar chain is involved in the biosynthesis of cell wall components [22]. Deletion of the gene encoding UGM (glfA), the UDP-Galactofuranose transporter (glfB), or the Galf-transferase GfsA in A. fumigatus, leads to the absence of the galactofuran side chains of galactomannan and is associated with increased susceptibility to antifungals, attenuated virulence, and reduced growth [16,19,20,24]. More recently, three putative Galftransferases, GfsA, GfsB and GfsC, were identified in A. niger. Analysis of the single, double and triple mutants revealed that GfsA together with GfsC are most important for the biosynthesis of Galf-containing sugars and multiple gfs gene deletions caused a severe cell wall stress response [24]. These reports reveal that a lack of Galf-containing sugar chains affects normal hyphal growth, cell wall structure and conidiation in filamentous fungi.
Fusarium oxysporum is a relevant plant pathogen causing Fusarium wilt syndrome in more than 120 different hosts and have been ranked in the top 10 plant fungal pathogens [25]. Although β1-6-linked Galf residues have been identified in the cell wall of Fusarium [26], the significance of Galf sugar residues in F. oxysporum remained to be unknown. Therefore, it is of important to evaluate the significance of Galf biosynthesis in growth and pathogenicity of this species.
In this study, BLASTp searches identified two putative UGM genes, ugmA (EGU88002.1) and ugmB (EGU83790), and seven putative Galf-transferase genes, EGU86692, EGU81169, EGU76406, EGU88219, EGU80311 and EGU82964 in the genome of F. oxysporum. The single and double mutant strains of the UGM genes, as well as the single mutant of the gfsB (EGU86692) were obtained in F. oxysporum f.sp. cucumerinum, a strain infecting cucumber and resulting in serious vascular disease worldwide [27]. By analyzing these mutants, the functions of Galf-containing sugars in growth, conidiation and pathogenicity of F. oxysporum f.sp. cucumerinum were examined.

Strains, plasmids, and culture conditions
F. oxysporum f.sp. cucumerinum (CGMCC3.2830, China General Microbiological Culture Collection Center) was used as the wild-type strain (WT) in this work. F. oxysporum f.sp. cucumerinum strains were cultured at 28˚C in potato dextrose agar (PDA) or in potato dextrose broth (PDB) with shaking at 200 rpm for production of fungal spores and mycelia. Conidia were washed with ddH 2 O from 7-days culture on plates and counted in a haemocytometer. Mycelia were harvested by filtering the 2-days cultures with two layers of slow qualitative filter paper. The pSCN44 vector harboring a hygromycin gene was used for mutant construction and the pKN vector carrying a neomycin gene was used for generating the double-mutant and the complemented strains. All vectors and plasmids were propagated in Escherichia coli DH5α.

Transformation, selection and characterization of stable transformants
The protoplasts were prepared according to Sarrocco et al. [28] with some modification. One milliliter of 10 9 ml -1 freshly harvested conidia of F. oxysporum f.sp. cucumerinum was inoculated in 100 ml of YPD medium and incubated at 28˚C, 200 rpm for 5-7 hours. The germinated conidia were collected by centrifugation at 3,000 rpm and washed three times with 0.7 M NaCl, then resuspended by gentle vortexing in an enzyme mixture containing lysing enzyme (10 mg ml -1 ), cellulase (15 mg ml -1 ), and driselase (10 mg ml -1 ) (Sigma-Aldrich) in 0.7 M NaCl and incubated at 30˚C, 100 rpm for about 1 hour. The protoplasts were counted by a hemocytometer and resuspended in STC buffer (1.2 M sorbitol, 10 mM Tris-HCl, pH8, 50 mM CaCl 2 ) in a final concentration of 10 8 CFU ml -1 . For the transformation, an aliquot of 200 μl protoplast suspension was gently mixed with 5 μg DNA and 50 μl of 30% PEG solution (30% PEG 8000, 10 mM Tris-HCl, pH8, 50 mM CaCl 2 ) and incubated at room temperature for 20 minutes, then 2 ml of 30% PEG solution were added, mixed carefully and incubated at room temperature for 5 minutes. Finally, 4 ml of STC buffer were added and mixed thoroughly. The transformants were screened using 50 μg ml -1 hygromycin B (Sigma-Aldrich) and screened again with 150 μg ml -1 hygromycin B. To obtain the double-deletion mutant, G418 was used to screen the mutant strains. For the complemented strains of the double mutants, 100 μg/ml phleomycin was used as the selection marker.

Targeted gene replacement and complementation
For the ΔugmA and ΔugmB mutant, the 1500-bp fragments of the upstream and downstream flanking sequences of the ugm genes were obtained by PCR amplification from wild-type genomic DNA using the primer pairs of ugmu1/ugmd1 and ugmu1/ugmd2 (S1 Table), respectively, and the hygromycin resistance gene (hyg) was amplified from the pCSN44 plasmid with the primer pair Hyg1/Hyg2. Then, the inserted cassette was generated by fusing the upstream and downstream flanking regions to the two ends of the hyg gene by overlapping-PCR. Finally, the cassette fragments were transformed into the protoplast of the wild-type strain. A double mutant was obtained by deletion of ugmA in the ΔugmB mutant. Complementation of the ΔugmA mutant was achieved by introducing pKN-ugmA, which was obtained by DNA amplification using primer pair ugmA C-F/ugmA C-R (S1 Table). For the completement of the double mutant, 100 μg/ml of phleomycin was used to screen the transformants. In all cases, HygR, G418R, or PhleoR resistant transformants were obtained and the homologous recombination or complementation events were confirmed by PCR and Southern blotting. The primers and probes used in this study are listed in S1 Table. KOD FX DNA polymerase (TOYOBO) was used in all PCR experiments. Plasmid DNA was extracted using the Plasmid Miniprep Kit (Axygen, China). Fungal genomic DNA was extracted using the method described previously [29]. Using the same method, the ΔgfsB mutant and its complemented strain was constructed using the primers shown in S1 Table. All mutants were confirmed by Southern Blotting.

Phenotypic analysis of the mutants
A droplet of 1×10 5 freshly harvested conidia was inoculated at the centre of a plate and cultured at 28˚C. The growth of the colonies was evaluated after 7 days. The conidia production of the strains were quantified in static cultures grown in continuous dark as described previously [30].
Germination analysis was carried out by inoculating 1×10 5 /ml or 1×10 6 /ml of freshly harvested conidia onto a glass slide with double concave in 50 μl PDB liquid medium. After incubation at 28˚C for specific times, the slides were taken out and observed [32].

Expression of the ugm genes in the WT and mutant strains
The freshly harvested conidia were cultured in PDB medium for 6, 12, 24, 36 and 48 hours. The hyphae were harvested by filtration through Whatman filter paper, washed with sterile water, and then ground under liquid nitrogen using a mortar and pestle. Total RNAs were isolated from ground conidia and mycelia using Trizol reagent (Invitrogen) according to the manufacturer's protocol. Complementary DNA synthesis was generated with 2 μg of RNA using the RevertAid First Strand cDNA Synthesis Kit (ThermoScientific). The primers (ugmA5'/ugmA3' and ugmB5'/ugmB3') used are listed in S1 Table. Whenever possible, the primers spanned an intron. Gene expression was analyzed with 300 nM primers in 20 μl reaction buffer by using EvaGreen qPCR Mastermix (ABM Company) and an iCycler apparatus (Bio-Rad). Cycle conditions were 10 min at 90˚C and 40 cycles of 15 s at 95˚C and 1 min at 60˚C. Single PCR products were confirmed with the heat dissociation protocol at the end of the PCR cycles. For each primer pair, the amplification efficiency was determined by serial dilution experiments, and the resulting efficiency coefficient was used for the quantification of the products. All reactions were performed in triplicate. The relative mRNA expressions were calculated according to 2 -ΔΔCt with the transcription elongation factor gene (EF) as the reference.

Detection of Galf with antibody
The anti-Galf monoclonal antibody AB135-8, which was identified and cloned from a pool of hybridoma cells obtained after immunization of Balb/c mice with killed A. fumigatus germ tubes [33]. Resting conidia were incubated in PDB liquid medium for 8 h and fixed with 2.5% p-formaldehyde (PFA) overnight at 4˚C. After fixation, cells were washed with 0.2 M glycine in PBS for 5 min, then sealed with 5% BSA in PBS for 1 h. Cells were incubated with the hybridoma culture supernatant containing AB135-8 at 1:10 dilution in 5% BSA in PBS for 1 h at room temperature. After washing with 1% BSA in PBS, cells were incubated with a goat Dylight 549-conjugated antibody directed against mouse IgG (H+L) diluted 1:2000 in BSA/ PBS. After washing with PBS, cells were visualized with an inverted fluorescence light microscope.

Virulence
The cucumber seeds (Chinese cultivar "Zhong Nong 16") were processed for accelerating germination as described by Li et al. [34], and the germinated seeds were grown in plastic pots filled with sterilized soil composed of vermiculite and turf. The pots were placed in the greenhouse at approximately 28˚C for a 16 h photoperiod until the two-true leaves stage. For virulence test of ΔugmA, ΔugmB and double mutant, the seedling roots were slightly wounded and the cucumber plants were cultured in 1 g/L Murashigeand Skoog medium (MS) with vitamin supplements containing 2×10 5 freshly harvested conidia. The pots were placed in green house with the same condition as described above. Twelve plants were used for each treatment. For virulence test of ΔgfsB, the slightly wounded seedling roots were immersed in conidia suspension of 5×10 6 /ml for 30 minutes. Subsequently, the infected seedlings were transplanted to plastic pots and placed in green house. Disease severity index (DSI) was calculated as previously described [35]  To examine the invasion of F. oxysporum f.sp. cucumerinum in plant vascular bundles, the root cross sections of infected plants were observed with a M205A stereoscopic microscope (Leica Microsystems).
To quantify the relative mycelial biomass of the fungal strains in cucumber roots, DNA samples were extracted from the cucumber roots cultured for 30 days after infection and subjected to quantitative PCR (qPCR) using primers EF1a-1/-2 for the transcription elongation factor gene of F. oxysporum f.sp. cucumerinum. The amounts of DNA templates were normalized using the cucumber actin 1 gene (primers: ACT1F/ACT1R). Three simultaneous replicated amplifications were carried out for each DNA sample, using 20-μL aliquots from a 60-μL mixture, and Bright Green qPCR Mastermix kit (ABM Company). qPCRs were performed in an iCycler apparatus (Bio-Rad) using the following cycling protocol: an initial step of denaturation at 95˚C for 10 min, followed by 40 cycles of 15 s at 95˚C, 60 s at 60˚C. After amplification, a melting curve program was run for which measurements were made for 5 min within a range of 65-95˚C. The DNA concentration of each sample was extrapolated from standard curves, which were developed by plotting the logarithm of known concentrations (10-fold dilution series from 100 ng to 1 ng/20 μL reaction) of F. oxysporum f.sp. cucumerinum genomic DNA against the Ct values. The relative DNA amounts of the indicated strains were calculated by comparative ΔCt from the mean of two different determinations of the threshold cycle. Each PCR reaction was followed by a melting curve, to assure that there was only one product amplified. All primers used in this assy are listed in S1 Table.
The ugmA, ugmB and gfsB single mutants were constructed as described in Materials and methods. Using hygromycin resistance as a selective marker, transformants of the ΔugmA, ΔugmB and ΔgfsB were screened. The complemented strains of ΔugmA were obtained with G418 sulfate screening. At the same time, the double mutant ΔugmAΔugmB was also constructed. The complemented strains ReumgA and ReumgB were obtained by re-introducing ugmA and ugmB into the double mutant. All mutant strains were verified by PCR and Southern blotting (S2 Fig).

Phenotypes of the mutants
As compared with the WT or complemented strains grown on PDA, the ΔugmA mutant showed a slightly reduced growth and the ΔugmB did not exhibit any difference, while the double mutant exhibited a severely retarded growth (Fig 1A). In contrast, the ΔgfsB mutant did not show any growth phenotype on PDA or MM (Fig 1B). After cultivation on PDA at 28˚C for 7 days, the ΔugmA mutant produced less conidia than the WT, while the double mutant produced only very few conidia (Fig 2A and 2B). The number of conidia produced by the ΔgfsB mutant was reduced by 28% ( Fig 2C). The germination of the ΔugmA and the double mutant was delayed for 2 h and 4 h, respectively (Fig 3A), while the germination of the ΔgfsB mutant was similar to that of the WT (Fig 3B). Theoretically, the phenotypes of the DM-ReugmB strain should be similar with the ΔugmA strain. However, in our study we found that complementation of the double mutant strain with the ugmB gene did not restore the growth and conidiation to the levels similar to the ΔugmA strain (Figs 1A and 2), which might be due to a lower activity of the plasmid we used for complementation of the double mutant.
On PDA medium supplemented with calcofluor white or Congo red, both ΔugmA and the double mutant showed increased sensitivities. The double mutant was more sensitive to calcofluor white and Congo red than the ΔugmA, while the sensitivity of the ΔugmB mutant was similar to that of the WT (Fig 4A). When the ugm mutants were incubated on MM medium, both ΔugmA and double mutant were hypersensitive to calcofluor white and Congo red. The sensitivity of the ΔugmA to calcofluor white and Congo red was able to be rescued by the addition of 1.2 M sorbitol, while the double mutant not. These results suggest that the cell wall defect of the double mutant is more severe than that of the ΔugmA. In contrast, the ΔgfsB mutant did not show any increased sensitivity to calcofluor white or Congo red (S3 Fig).
These results suggest that the ugmA contibutes to the cell wall integrity, hyphal growth and conidiation, whereas the gfsB is not required.

Synthesis of Galf-containing sugar chains in the mutants
As the ΔugmB mutant did not showed any phenotypic changes and the ΔugmA mutant exhibited defects in growth, cell wall integrity and conidiation as compared with the WT, it is likely

PLOS ONE
that the ugmA can compensate for the loss of the ugmB gene. To testify this hypothesis, we analyzed the expression of the ugmA and ugmB genes in the WT and mutant strains. RT-PCR analysis revealed that the ugmA was expressed during hyphal growth (6-24 h) and highest expression occurred at log-phase (12 h), while the ugmB was expressed during conidiation (after 24 h) in the WT (Fig 5A), which suggests that UgmA contributes to the hyphal growth whereas UgmB contributes to the formation of conidiospores. In the ΔugmB mutant the expression of the ugmA was increased during hyphal growth and conidiation with the highest expression level at 24 h (Fig 5B), while the expression of the ugmB was remarkably increased (0.2-10 folds) in the ΔugmA mutant ( Fig 5C). As the ΔugmB mutant did not show any phenotype, it is reasonable to conclude that UgmA can completely compensate for the loss of UgmB, while UgmB could only partially complement the function of UgmA. Our results indicate that UgmA is more important for Galf-containing sugar synthesis during vegetative growth of F. oxysporum f. sp. cucumerinum.
As shown in Fig 6, when the mutants were detected with the galactomannan-specific antibody AB135-8 that specifically recognizes galactofuranose [33], weak signals were detected

Virulence of the mutants
To evaluate the significance of the ugmA, ugmB and gfsB genes for the virulence of F. oxysporum f. sp. cucumerinum, 2×10 5 conidia/ml of each strain were inoculated to the cucumber seedling roots as described in Materials and methods. The infected seedlings were cultured at

PLOS ONE
28˚C in green house. Apparently, as shown in Fig 7A, the cucumber seedlings infected by the ΔugmA and double-deletion mutant showed a slight wilt syndrome, indicating that the virulence of these two mutants was greatly attenuated. Calculation of the disease severity index (DSI) [35] also revealed that the DSI of both ΔugmA and double mutant were 25% as compared with 70.8% for the WT and 18.7% for mock infected plants, while the virulence of the ΔugmB mutant was similar to the WT (Fig 7E). These results indicated that the ugmA has a direct function in virulence of F. oxysporum f. sp. cucumerinum.
F. oxysporum f. sp. cucumerinum infects the plants through the roots. As shown in Fig 7B, the main roots from the plants infected by the WT strain and ΔugmB strains showed brown color, a rotten appearance and fewer lateral roots, while the main roots infected by the ΔugmA and double mutants suffered much less than those of the WT. Examination of root sections under a stereoscopic microscope (Fig 7C) and qPCR analysis (Fig 7D) revealed that the vascular bundles of roots that were infected with the ΔugmA or the double mutant were comparable to mock infected roots, while those infected by the WT and the ΔugmB strain were severely brown stained in their cortex and vascular bundles. These results demonstrate that ugmA was required for the virulence of F. oxysporum f. sp. cucumerinum in cucumber. Interestingly, the ΔgfsB mutant also exhibit a reduced virulence although no visible phenotypes were detected in the other assays (Fig 7A-7C and 7E).

Discussion
Fusarium species are common plant pathogens [25], but some species are also pathogenic for humans, particularly F. solani, F. oxysporum, and F. moniliforme, causing superficial infections in immnocompetent individuals, and severe and disseminated disease in immunocompromised hosts [36,37]. F. oxysporum Galf-containing sugar chain consists of a main chain containing β1,6-linked Galf residues with side chains containing α1,2-linked Glcp, β1,2-linked Manp and β-Manp terminal nonreducing end units, which is different Fig 7. Virulence of the mutant strains. In A and B, the germinated cucumber seeds in plastic pots were grown in the greenhouse at approximately 28˚C for a 16 h photoperiod until the two-true leaves stage. The infected seedlings were analyzed after 45 days. Twelve plants were used for each treatment. In C, the root cross sections of infected plants were observed with M205A stereoscopic microscope (Leica Microsystems) (×20). In D, to quantify the relative mycelial biomass of the fungal strains in cucumber roots, DNAs were extracted from the cucumber roots cultured for 30 days after infection and subjected to quantitative PCR (qPCR) using primers EF1a-1/-2 for the transcription elongation factor gene of F. oxysporum as described in Materials and methods. In E, disease severity index (DSI) was calculated as previously described from 2 weeks of inoculation up to 45 days. The plant infection experiments were repeated for three times. https://doi.org/10.1371/journal.pone.0250064.g007

PLOS ONE
from that of A. fumigatus and important for the activation of macrophage mechanisms [26]. As Galf is not found in mammalian and plant cells [12], inhibition of the biosynthetic pathway of Galf-containing sugar chains is an ideal target for the development of medicines and agricultural chemicals, which can specifically inhibit F. oxysporum without any side effects.
The genes encoding the UDP-Galp mutase have recently been identified in several fungal species such as Cryptococcus neoformans, A. fumigatus, A. nidulans and A. niger [10,14,21]. In A. fumigatus, abrogation of Galf-synthesis causes attenuated virulence and reduced growth [16,19,20,23], indicating that a lack of Galf-containing sugar chains affects normal hyphal growth, cell wall structure and conidiation in filamentous fungi.
In this report, we used F. oxysporum f.sp. cucumerinum as a model species to evaluate the biologial functions of Galf-containing sugar chain in F. oxysporum. Two UGM genes (ugmA and ugmB) were identified in the genome of F. oxysporum. Phentypes of the single and double mutant of the UGM genes were investigated. Based our results, UgmA was required for vegetative growth and UgmB contributed to conidiation in F. oxysporum f.sp. cucumerinum. Although the ugmB was highly expressed in the ΔugmA, only a weak signal was detected by antibody AB135-8 on the cell wall of the ΔugmA mutant (Figs 5C and 6), suggesting that the residual Galf-containing sugars were not enough for maintaining the mycelial cell wall integrity. On the other hand, the elevated expression of the ugmA could compensate for the loss of the ugmB (Fig 5B). These results suggest that UgmA is able to complement the function of UgmB during conidiation, whereas UgmB is not able to function as UgmA during vegetaive growth stage. Although the mechanism remains unclear, it is likely that UgmA and UgmB differ in their activities. Further investigation should be carried out to characterize these two enzymes.
We found that both ΔugmA and double mutant showed an attenuated virulence, suggesting UGMs, especially UgmA, play an important role in virulence. Therefore, UgmA represents an attractive target for the development of new drugs.
In A. fumigatus, the cell wall galactomannan (GM) is composed of a main α1,2/α1,6-mannan chain and a number of side chains of β1,5/β1,6-linked Galf residues in its cell wall [3]. Eight putative gfs genes are identified in the genome of A. fumigatus, including gfsA (Afu6g02120), gfsB (Afu4g13710), gfsC (Afu4g10170), gfsD (Afu6g00520), gfsE (Afu3g07220), gfsF (Afu3g00170), gfsG (Afu2g17320) and gfsH (Afu8g07260). All eight Gfs proteins can be phylogenetically classified into two groups: GT31-A and GT31-B [6]. Recently, GfsA is biochemically characterized as the enzyme responsible for the transfer of β1,5-linked Galf to the side chain of GM [20]. As GfsA, GfsB and GfsC are belonged to GT31-A family, GfsB and GfsC are also predicted as β1,5-galactofuranosyltransferases similar to GfsA [20]. GfsA, GfsB and GfsC are also identified in the genome of A. niger and A. nidulans [19,24]. In contrast to A. fumigatus, A. niger and A. nidulans, no GfsA homolog was found in the genome of F. oxysporum (Fig 8), suggesting that the linkage of cell wall Galf residues are different in F. oxysporum. Indeed, it has been reported that Fusarium produces cell wall β1-6-linked Galf residues [26]. Therefore, it is not surprised that we were unbable to find GfsA homolog in F. oxysporum. We only found two GT31-A family members in F. oxysporum, GfsB (EGU86692) and GfsC (EGU81169), which share 35% and 25% of identity with A. fumigatus GfsB and GfsC, respectively (Table 1 and Fig 8). In this study, the significance of the gfsB gene was evaluated. Deletion of the gfsB gene did not show any visible phenotype, however, the ΔgfsB mutant displayed a reduced virulence. Although its substrate specificity needs to be characterized, it is no doubt that GfsB is also an ideal drug target for the development of novel drugs against F. oxysporum.   Table. Oligonucleotides used in this study. (PDF)